JMIR Publications

ABSTRACT

Background:

Longitudinal cohort studies are critical for understanding if and how behaviors that influence health, such as e-cigarette use, evolve over time. Optimizing follow-up rates in longitudinal studies is necessary for ensuring high quality data with sufficient power for analyses. However, achieving high rates of follow-up in web-based longitudinal studies can be challenging, even when monetary incentives are provided.

Objective:

We compare participant progress through the survey and demographics for two incentive delivery structures (conditional, hybrid unconditional-conditional) among our sample of U.S. adults who use e-cigarettes to understand the optimal incentive structure.

Methods:

Data are from a web-based longitudinal cohort study (Wave 4; July-Sept 2022) of U.S. adults (≥21 years) who use e-cigarettes ≥5 days/week. Participants (n=1804) invited to the follow-up survey (median completion time: 16 minutes) were randomly assigned into one of two groups, each offered a different incentive structure: (1) Conditional: $30 gift code upon completion of the survey; (2) Hybrid unconditional-conditional: $15 gift code prior to completion and $15 gift code upon completion of the survey (n=902 each). Chi-squared tests were used to assess the difference in participant progress through five sequential stages of the survey (started survey, completed screener, deemed eligible, completed screener, deemed valid) and demographics between these two groups.

Results:

Of the 902 participants invited to the follow-up survey in each group, a higher proportion of those in the conditional (n=662; 73.4%) than the hybrid (n=565; 62.6%) group started the survey (P<.001). Of those who started the survey, 643 (97.1%) in the conditional group and 548 (97.0%) in the hybrid group completed the screener (P=.89), employed each wave to ensure participants remained eligible. Of those who completed the screener, 555 (86.3%) in the conditional group and 446 (81.4%) in the hybrid group were deemed eligible for the survey (P=.02). Of those eligible, 514 (92.6%) from the conditional group and 401 (89.9%) from the hybrid group completed the survey and were deemed valid after a review of the data (P=.14). Overall, more valid completions were yielded from the conditional (n=514; 57.0%) than hybrid group (n=401; 44.5%) (P<.001). Among those who validly completed the survey, no significant differences were found by group for gender, income, race, ethnicity, region, e-cigarette use frequency, past 30-day cigarette use, or number of waves previously completed.

Conclusions:

Providing a $30 gift code upon completion of the survey yielded higher rates of survey starts and completions than providing a $15 gift code both before and after survey completion. These two methods yielded participants with similar demographics, suggesting one approach is not superior in obtaining a balanced sample. Based on this case study, future web-based surveys examining U.S. adults who use e-cigarettes could consider providing the full incentive upon completion of the survey.